Interesting idea, but I don't think the problem he posits is new. Everything is done along a chain of trust. Even current researchers don't reproduce every experiment that led up to their current body of knowledge. No one has ever understood the entire chain of reasoning.
More to the point, mistaken ideas have always infected lots of people, this was the impetus for science in the first place. This will resolve itself in the usual way- people rely on the results, the results turn out to be wrong, and we search for more answers.
I wonder how long until the trust chain is automated? As long as the results proceed along the lines of logical deduction and statistics, it can be machine understandable.
